Not every rejection is about the image. A large share of stock rejections are triggered by the metadata — a brand name in the keywords, a missing release implied by a term like “model”, or commercial metadata on an editorial shot. These are the cheapest rejections to prevent, because you can catch every one of them before you ever hit upload.
The 7 triggers, in order of how often they bite
1. Trademarks and brand names in keywords
"iPhone", "Nike", "Tesla", "Lego" — even if the product is unrecognizable in the frame, the keyword alone can trigger an intellectual-property rejection on commercial content. Describe the object generically: smartphone, sneakers, electric car.
2. Terms that imply a missing release
Keywords like "model", a person's name, or a recognizable private location tell the reviewer a release should exist. If you can't attach a model or property release, the metadata shouldn't promise one.
3. Commercial metadata on editorial content
News events, logos in frame, identifiable people without releases — that's editorial. Submitting it with commercial-style metadata (and no editorial caption format) is an instant rejection on most platforms.
4. Keywords not visible in the image
Adding "summer" to a studio shot or "family" to a photo of one person is keyword spam to a reviewer. Every term must be defensible by pointing at the frame.
5. Repeated or near-duplicate keywords
"dog, dogs, doggy, puppy, puppies" wastes slots and reads as stuffing. Agencies' quality filters score this pattern automatically.
6. Identical metadata across a whole batch
Copy-pasting one keyword block onto 50 different images is the most machine-detectable spam pattern there is. Similar images may pass review individually and still get the whole batch flagged.
7. AI-generated content not declared
Adobe Stock and others require generative AI content to be labeled as such — and reject it when the metadata (or the title) pretends it's a photograph. Declare it, and keyword it honestly.
Why rejections cost more than the lost upload
A rejection isn't just one image bounced. Review queues on major platforms run from days to weeks, so a rejected batch means re-editing metadata and going to the back of the line — and repeat offenders see their whole account's review priority drop. Ten minutes of metadata checking protects weeks of queue time.
Avoid metadata triggers from the start
All seven triggers above are text patterns — and clean metadata avoids most of them by design. AI Keyword Genius writes titles, descriptions and keywords from what is actually visible in the frame: literal, descriptive terms instead of brand names, no release-implying claims and no duplicate keyword stuffing. Pair that with the checklist below and metadata stops being the reason a file bounces.
A 60-second pre-upload checklist
- ✅Search your keyword list for brand names — replace with generic terms
- ✅If a person or private property is recognizable: release attached, or route it as editorial
- ✅Delete any keyword you can't point to in the frame
- ✅Deduplicate: one form per concept (dog, not dog + dogs + puppy)
- ✅Confirm no two images in the batch share an identical title or keyword block
- ✅Generative AI content: declared, and titled as an illustration/render